Hi Ramsey,
Thanks for the tips and do agree that is one of ways to work out.
I would like to chat with you a puzzle thing that bothers me. As we knew to use dsplink is to speed up some code in dsp core. However, my test code shows that is not a case.
I use scale sample and add time to measure the time to scale 1024 float data in dsp and gpp(my process is omapl138).
The starting point of measurement from calling SCALE_Execute () at SCALE_Main() in scale.c and ending point of measurement is after SCALE_Execute() called , which takes 3392 us. The gpp does same scale data only takes 471 us.
Any idea why ggp processes faster than dsp core?
Thank you for taking look.
Joe